From 4c52b7fcc694ebcbc9db5039759e30a73c7d5498 Mon Sep 17 00:00:00 2001 From: Shangming Cai Date: Thu, 19 Mar 2026 19:12:50 +0800 Subject: [PATCH] [CI] Improve PP consistency check success rate (#20838) Signed-off-by: Shangming Cai --- .../distributed/test_pp_single_node.py | 18 +++++++++--------- 1 file changed, 9 insertions(+), 9 deletions(-) diff --git a/test/registered/distributed/test_pp_single_node.py b/test/registered/distributed/test_pp_single_node.py index a051764f7..2468b82b7 100644 --- a/test/registered/distributed/test_pp_single_node.py +++ b/test/registered/distributed/test_pp_single_node.py @@ -31,7 +31,7 @@ from sglang.test.test_utils import ( run_bench_one_batch_server, ) -register_cuda_ci(est_time=500, suite="stage-c-test-4-gpu-h100") +register_cuda_ci(est_time=650, suite="stage-c-test-4-gpu-h100") class TestPPAccuracy(unittest.TestCase): @@ -214,7 +214,7 @@ class TestQwenPPAccuracy(unittest.TestCase): args = SimpleNamespace( num_shots=5, data_path=None, - num_questions=200, + num_questions=512, max_new_tokens=512, parallel=128, host="http://127.0.0.1", @@ -238,7 +238,7 @@ class TestQwenPPAccuracy(unittest.TestCase): pp_metrics["accuracy"], baseline["accuracy"] - 0.02, msg=( - f"PP accuracy dropped more than 1% compared to baseline. " + f"PP accuracy dropped more than 2% compared to baseline. " f"Baseline: {baseline['accuracy']:.2%}, PP: {pp_metrics['accuracy']:.2%}" ), ) @@ -269,7 +269,7 @@ class TestQwenPPTieWeightsAccuracy(unittest.TestCase): args = SimpleNamespace( num_shots=5, data_path=None, - num_questions=200, + num_questions=512, max_new_tokens=512, parallel=128, host="http://127.0.0.1", @@ -292,7 +292,7 @@ class TestQwenPPTieWeightsAccuracy(unittest.TestCase): pp_metrics["accuracy"], baseline["accuracy"] - 0.02, msg=( - f"PP accuracy dropped more than 1% compared to baseline. " + f"PP accuracy dropped more than 2% compared to baseline. " f"Baseline: {baseline['accuracy']:.2%}, PP: {pp_metrics['accuracy']:.2%}" ), ) @@ -321,7 +321,7 @@ class TestQwenMoePPAccuracy(unittest.TestCase): args = SimpleNamespace( num_shots=5, data_path=None, - num_questions=200, + num_questions=512, max_new_tokens=512, parallel=128, host="http://127.0.0.1", @@ -344,7 +344,7 @@ class TestQwenMoePPAccuracy(unittest.TestCase): pp_metrics["accuracy"], baseline["accuracy"] - 0.02, msg=( - f"PP accuracy dropped more than 1% compared to baseline. " + f"PP accuracy dropped more than 2% compared to baseline. " f"Baseline: {baseline['accuracy']:.2%}, PP: {pp_metrics['accuracy']:.2%}" ), ) @@ -375,7 +375,7 @@ class TestQwen35PPAccuracy(unittest.TestCase): args = SimpleNamespace( num_shots=5, data_path=None, - num_questions=200, + num_questions=512, max_new_tokens=512, parallel=128, host="http://127.0.0.1", @@ -398,7 +398,7 @@ class TestQwen35PPAccuracy(unittest.TestCase): pp_metrics["accuracy"], baseline["accuracy"] - 0.02, msg=( - f"PP accuracy dropped more than 1% compared to baseline. " + f"PP accuracy dropped more than 2% compared to baseline. " f"Baseline: {baseline['accuracy']:.2%}, PP: {pp_metrics['accuracy']:.2%}" ), )